Reinforcement Learning in Autism Spectrum Disorder
Signe Bray
University of Calgary

Autism spectrum disorder (ASD) is a complex and heterogeneous neurodevelopmental disorder diagnosed in early childhood. Children with ASD show challenges in social communication, repetitive motor behaviors and restricted, idiosyncratic interests. One approach to mitigating these challenges is through behavioral interventions that use repeated practice with reinforcement. However, several recent lines of evidence suggest that the neurocircuitry supporting the brain’s ability to learn new associations may function atypically in ASD. A better understanding of how the learning brain is different in children with ASD could help to refine and improve early therapies. We have therefore undertaken a series of studies examining how youth with ASD respond to rewards and learn from feedback, and asked whether there are differences in the structure of brain regions important for learning. We find that there are subtle structural alterations in the thalamus and striatum, regions that are critical for associative learning. However, under certain conditions, namely when reinforcers are tailored to an individual’s interests, behavioral and neural response to rewards is largely typical. I will discuss our findings in the context of social motivation theories of ASD, inter- and intra-individual variability and implications for therapeutic interventions.